US5521985A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Apparatus for recognizing machine generated or handprinted text
Assignee: International Business Machines Corp
•
Filing: 1994-11-28
•
Granted: 1996-05-28
•
Expires: 2014-11-28
Abstract
Abstract A computer system with an image reader and nerual network is provided for determining whether an image of text has been generated by a machine or by hand. This serves the useful purpose of allowing one to use speciallized recognition techiques that are more suited to one form of printing, thus achieving a higher recognition accuracy than by using a single recognition technique for both types of printing. The method is based on the premise that the spatial spectra for an image of machine text will have more higher frequency components than one generated by hand, because of the nonregular, nonuniform slant of the handprint. The method proposed generates this spectra by convolving spatial templates with vertical histograms from each line of text, and uses a neural network to classify the resulting spectra.
Inventors
- 👤 William O. Camp, Jr.
- 👤 Keith J. Werkman
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-10 00:50
CPC Classifications
G; G06; G06V; G06V30/00; G06V30/10; G06V30/24; G06V30/242; G06V30/244; G06V30/2455
Claims Analysis
Total Claims: 4
Claim 1
Independent Claim
273 words
An apparatus for recognizing whether text in an image is machine print or hand print, the apparatus comprising: a feed forward neural network processor for processing digital information; a memory for storing digital information processed by said feed forward neural network processor, an image reader for reading an image containing text into the memory in pixel format; a means for separating the image into horizontal lines of text, using a horizontal histogram of pixel counts; a means for processing each said horizontal line of text into a vertical histogram; a convolving means for convolving across the vertical histograms with each of a plurality of templates, each template representing a spatial filter, accumulating the responses for each template from each horizontal position of the vertical histograms as each template is convolved across the vertical histograms, the accumulated responses from each of the templates collectively forming a spectra having high frequency components and low frequency components and whose characteristic is dependent on whether it is machine printed or hand printed; and the feed forward neural network processor looking for the relative relationship of high frequency to low frequency components for classifying the spectra as machine print or hand print, said neural network having the spectra as input and weight means for multiplying the spectra input at nodes and accumulating a final output thereof at final output nodes, with a first output providing an indication that the text contained in the image is machine print, with high frequencies having high relative amplitude, and a second output providing an indication that the text contained in the image is hand print with low frequencies having high relative amplitudes.
Claim 2
Independent Claim
273 words
An apparatus for recognizing whether text in an image is machine print or hand print, the apparatus comprising: a feed forward neural network processor for processing digital information; a memory for storing digital information processed by said feed forward neural network processor, an image reader for reading an image containing text into the memory in pixel format; a means for separating the image into horizontal lines of text, using a horizontal histogram of pixel counts; a means for processing each said horizontal line of text into a vertical histogram; a convolving means for convolving across the vertical histograms with each of a plurality of templates, each template representing a spatial filter, accumulating the responses for each template from each horizontal position of the vertical histograms as each template is convolved across the vertical histograms, the accumulated responses from each of the templates collectively forming a spectra having high frequency components and low frequency components and whose characteristic is dependent on whether it is machine printed or hand printed; and the feed forward neural network processor looking for the relative relationship of high frequency to low frequency components for classifying the spectra as machine print or hand print, said neural network having the spectra as input and weight means for multiplying the spectra input at nodes and accumulating a final output thereof at final output nodes, with a first output providing an indication that the text contained in the image is machine print, with high frequencies having high relative amplitude, and a second output providing an indication that the text contained in the image is hand print with low frequencies having high relative amplitudes.
Claim 3
Dependent Claim
54 words
An apparatus according to claim 1 wherein the spectra formed has high frequency components and low frequency components wherein for machine print the high frequency components have a higher relative amplitude than the low frequency components and for hand print the low frequency components have a higher relative amplitude than the high frequency components.
Claim 4
Dependent Claim
29 words
An apparatus according to claim 2 wherein the neural network process uses the relative amplitudes of the high frequency components and the low frequency components to classify the spectra.
References Cited (Backward Citations)
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1999-05-28 | 1999-11-27 | 2000-05-28 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2003-05-28 | 2003-11-27 | 2004-05-28 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2007-05-28 | 2007-11-27 | 2008-05-28 | $7700.00 | Window Closed / Lapsed |